See No Evil, Hear No Evil, Stop No Evil: Institutional-Level Tracking to Combat Mistreatment of Residents and Fellows

Verbal abuse, sexual harassment, and discrimination are encountered by residents and fellows across multiple specialties.^1–6  Types of mistreatment previously identified in the literature include offensive remarks, microaggressions (ie, brief everyday exchanges that send denigrating messages to certain individuals because of their group membership),¹  threats of physical harm, humiliation or shame, and sexual harassment.³  Teaching hospitals are complex workplaces, and trainees may be targets of mistreatment from multiple sources. These sources—attending physicians, consultants, colleagues, interprofessional team members, and patients—have different supervisory structures, and therefore different incentive mechanisms to modify their behavior.^7–9  Mistreatment is linked to decreased job satisfaction, burnout, attrition from the profession, and poor patient care.^3,10–12  Sources and prevalence of mistreatment vary by specialty and by trainee demographics.^2,3  However, a given source of mistreatment potentially interacts with multiple training programs. Most published efforts to track trainee experiences of mistreatment come from individual training programs, not sponsoring institutions.²  Trainee perceptions of mistreatment differ from medical student perceptions, but comparatively little attention has been paid to their experiences.¹³ 

Most institutional reporting mechanisms ask participants to identify individual perpetrators. Due to fear of retribution and concern for individual punitive action rather than restorative justice measures, victims significantly underreport mistreatment under such mechanisms.⁷  In contrast, climate surveys, which allow trainees to report perceptions of mistreatment anonymously, reveal a higher prevalence of mistreatment.¹⁴  However, climate survey results may be difficult to interpret and act upon with specificity.^14,15 

The current array of national surveys administered to trainees is inadequate for defining the scope of the problem and for regular tracking. The Association of American Medical Colleges (AAMC) Graduation Questionnaire (GQ) asks about discrimination, harassment, and shaming, but it lumps faculty, interprofessional staff, and other students into a single category of perpetrator(s), and it notably omits patients as a source of mistreatment.¹⁶  The Accreditation Council for Graduate Medical Education (ACGME) annual program surveys do not currently inquire about discrimination or harassment. Existing surveys either lump all sources of mistreatment together or ask only about a single source, usually a direct supervisor.^7,17  There are no tools, to our knowledge, that concurrently track both the range and sources of mistreatment of trainees at the institutional level.

Failure to specify sources of mistreatment precludes the institution from gaining an understanding of specific problem areas or meaningfully prioritizing resources to address problems. Thus, our goal was to adapt and administer a short institutional questionnaire that simultaneously tracks the range of mistreatment by varying sources.

The setting for this study was a large Midwestern medical school that sponsors 91 ACGME-accredited and 41 non–ACGME-accredited residency and fellowship programs. This study involved trainees who graduated in academic year 2017–2018.

Within the annual program evaluation survey administered to trainees, we included 6 questions focused on the mistreatment categories (figure). Five were modified versions of selected mistreatment questions from the AAMC GQ,¹⁸  and one was modified from the Racial Microaggressions Scale.¹⁹  A key modification was to ask trainees to report mistreatment from multiple sources. The survey was anonymous but gathered demographic information, including training program, age, race, and sex. The racial categories were taken from the US Census Bureau.²⁰ 

Items were piloted to ensure instructions, scales, and anchors were clear.²¹  Our pilot study indicated the questions took approximately 5 minutes to complete. The survey was introduced to residents and program directors by our associate dean for graduate medical education during monthly Graduate Medical Education Council meetings and through e-mail communications. Subsequently, we administered the survey using Qualtrics software to all graduating trainees in 2017–2018 and allowed 4 weeks to respond.

Although our scale had 3 anchors, data were heavily positively skewed, so we created 2 categories for each behavior: either never experienced or experienced at least once. For each behavior-source combination, we calculated the percentage of graduating trainees who experienced the behavior at least once in the previous academic year (table). Based on these results, we conducted additional analyses to determine whether trainees experienced higher rates of mistreatment from patients than from other sources. For each category of mistreatment, we conducted paired-sample t tests for each patient-other group combination. All comparisons were significant at P = .05 after making Bonferroni corrections for multiple comparisons. Data were analyzed using SPSS Statistics 24 (IBM Corp, Armonk, NY) by the director for graduate medical education, who holds a PhD in industrial and organizational psychology.

This study was determined to be exempt by the University of Minnesota–Twin Cities Institutional Review Board.

A total of 77% of trainees surveyed responded (234 of 304). A total of 53% (80 of 152) of respondents identified their sex as male, and 28% (63 of 222), 16% (36 of 222), and 11% (25 of 222) were third-, fourth-, and fifth-year residents, respectively. A total of 14% (31 of 222), 8% (17 of 222), and 13% (28 of 222) of our sample were first-, second-, and third-year fellows, respectively. The largest percentage of respondents came from our pediatrics (9%, 20 of 218), internal medicine (5%, 11 of 218), and surgery (4%, 8 of 218) residency programs. There was no cost for administration, as the survey was administered using software licensed by the University of Minnesota. The questions took less than 5 minutes to complete.

For every category of mistreatment except humiliation and shame, trainees experienced higher frequencies of mistreatment from patients than from any other source (table). Paired t test comparisons for patients versus all other sources, across mistreatment types, ranged from t = 3.92 to t = 9.71, all P < .001. Some departments had a higher prevalence of mistreatment by faculty. For example, in 1 department with 40 trainees, no trainees reported being the recipient of microaggressions from faculty in their program, whereas in another department with 31 trainees, 23% (7 of 31) of trainees reported being the recipient of this form of mistreatment from faculty in their program.

Our results show that patients were the primary source of mistreatment of graduating trainees. Further, our results demonstrate that this brief survey was feasible to administer, and the response rate belies an acceptability to trainees.

Our finding that patients are the primary source of mistreatment of trainees is consistent with a qualitative analysis employed by resident focus groups at one of our training sites.²²  In addition, other training programs have found that patients are a prevalent source of mistreatment of trainees.²  Collecting data across training programs in the institution allowed identification of institution-wide trends, including the universal prevalence of mistreatment by patients. However, by also identifying the training program of respondents, departmental and program-level trends could also be identified. For example, we found that some departments had higher prevalence of faculty mistreatment.

An important lesson learned from this pilot study is that, in order to inform interventions that are appropriately targeted, it may be desirable to obtain more nuanced data regarding frequency of mistreatment. Another lesson learned is that, in order to understand how mistreatment varies by important demographic variables, such as race, it is important to have very large sample sizes to conduct appropriate racial subgroup analyses. We recognize that other institutions may not have the same relationship between the sponsoring institution and the training programs, and that a survey such as ours, if seen as a mechanism to penalize programs, likely would have lower acceptability.

These results have prompted us to develop a more formalized institutional plan to address mistreatment. In particular, we are working on streamlining our approach to reporting patient misconduct and identifying institutional mechanisms to support and protect trainees when they experience mistreatment from patients. While our data show that we have work to do to reduce and mitigate mistreatment from all sources, understanding the most prevalent source of these behaviors provides a starting point for meaningful intervention.

Six questions concerning the types and sources of trainee mistreatment, embedded within an institutional survey, generated new information for institutional-, departmental-, and program-based future interventions. Patients were the greatest source for all types of mistreatment except humiliation and shaming.